Si2k And Red Cohabitation

[Bob Hart]

Here are a few still-images of a camera setup check for a musical performance to be covered in a few days time in Fremantle Western Australia.

The camera mix is two REDs to shoot 3K, one SI2K to shoot 2K and 2 JVC GY-HD100s to do the best they can as the close-up cameras either side of stage. Final intended output may be to HD and SD broadcast and SD DVD-Video.

1. SI2K and RED.
2. Steve Rice with his SI2K.
3. Wrangling the REDS.

It is believed there are now four REDs in Western Australia and one S12K that we know of.

[Chris Lakes]

Hey Bob,

so how did the shoot go? Any stills? Thanks.

[Bob Hart]

The first one was a bit of a frustration. My assigned camera had a battery problem of the JVC kind. Five batteries all fully charged, three newies and they all did not add up to the required two hours. Part of the problem was somebody unplugged the charger.

Last week the frame buffer on the SI2K was becoming consumed at full 2K so we had to down the resolution to the broadcast standard of 1920 x 1080 the footage will go out as.

We have been using P+S Technik's demonstrator camera for a fair while now so I think we are a few software revisions behind.

As for the JVCs, I am accustomed to Sonys and groundglass adaptors so am not fluent with the JVCs as yet.

Tonight was a better adventure, a bit of rain and a late start but the music, Lucky Oceans hosting and interviewing and backing on a pedal steel. --------- magic.

Tonight, I powered the JVC off mains and brought a HDTV along for better focusing. As well I did because there was a constellation of hot pixels staring me in the face that I did not spot. Fortunatedly my assistant did and Rob Castaglione did his good works in the special menu to make them go away.

I can't speak for the REDs and the SI2K as the post people take the footage away to do their thing to. The JVCs are in the wing positions and chasing the close-ups where resolution of the finer detail of wide shots is not an issue.

They might not be REDs or SI2Ks, but the images look good on the HDTV anyway. Motes of dust in the air, insects, hair strands, all there to be seen.

Because it is a live gig and the paying audience rightly has priority, there is a limit to how much we can intrude, so things like music stands and mike booms get in the shot from time to time and being tied to power and my old Miller slow-release tripod limits the options on my side of the stage somewhat.

The next one is going to be in a different venue so many of the hard learned lessons will have to be relearned next week. Should be a good gig.

I can't post any grabs as the footage is not in my hands. Hopefully John, or Rob might contribute if there are no copyright issues to get in the way and time permits.

[Chris Lakes]

Cool. Thanks for the report.

Yeah, you must have had outdated software on the SI-2K, as I've shot in very low light situations at full 2K 16:9 and had zero RAM buffer overruns. I was fully expecting that to be an issue in low light, but it works wonderfully. I had RAM overrun issues in the past, on earlier builds with the Mini, but not the full SI-2K. The laptop is actually faster, my problems were due to overheating the laptop before we realized it was an issue and took steps to keep it cool.

[Bob Hart]

I found on the first night that we had to come back one step below 2K and even then the buffer was slowly being consumed.

The camera was on the audience left corner back row and variously being used for full stage wides, two shots and medium close views.

One RED was doing the same from the other side. The centre RED was locked off master wide.

Interestingly, also discovered that if the frame was allowed to "rest" at a wide view without zooms, close-ups and follows it would recover slowly and then the gymnastics could begin again.

At 1920 x 1080 25P no problems at all. The meter barely moves.

[Chris Lakes]

probably working less hard on those shots. I used to have buffer overruns on the Mini when at a high resolution and shooting very dark scenes. Probably old software. The full cam I have has no issues, even in dark scenes with lots going on.

[Rohan Dadswell]

Hi Chris,

What build of software are you running ?
I've got 1.0.584 on my unit and I'm still having some buffer issues with 2K 25fps QT & AVI (especially with QT but that was to be expected)
I know that there is meant to be a new release soon - just looking for some conformation that this could ease the problem.

cheers
Rohan

[Chris Lakes]

The software build isn't brand new, just more recent than when my Mini was running into that issue under certain conditions. It has actually been a non-issue for some time now on both my Mini and SI-2K.

Which are you having the overruns with, Mini plus laptop or body? Are you forcing quality 4? What conditions do you shoot in that cause overruns? I'll try to remember to check tomorrow what build I'm running. I don't record QT, so I don't know how it performs compared to AVI.

The only difference that is apparent is you're shooting 25 as opposed to 24, which I guess could make a difference.

Oh, also, in Camera Preferences, under Display Mode, is the Rec/Play set to Bilin, Quad, or Hex? I usually set Preview to Bilinear, but Rec/Play to Quad or Hex (having already grabbed focus marks). I seem to remember that making a big difference on the last shoot. Quad is usually fine for me, and still not too soft.

Having asked whether you pin compression to quality 4, I have to say that I've tried doing that and didn't run into any problems. However, I tend to set it to adaptive as I'd rather not lose a shot in the middle of a take.

Hope you can figure it out.

Best,
Chris.

[Chris Lakes]

that reminds me ... I wonder why you can choose Bilin, Quad, or Hex for Preview, but Rec and Play are grouped together. Wouldn't playback need less RAM buffering than recording. I would think you could group Play with Preview, or separately. Or do I not understand something?

[Rohan Dadswell]

I did put in a feature request for the playback to be separated from the record as that always seemed odd to me as well - hopefully that will be in the update.

I've got the full body set up, haven't used the laptop since I received it.
I usually shoot adaptive (mainly out of dread of the buffer overrun - I'd prefer to lock the quality to 3 or 4)
Have just done a quick test at 24fps AVI and all seemed good so I guess the extra 4% of data is enough start filling the buffer.

Have tried several different hard drives which did make a difference - 7200rpm with a larger buffer.

Bob
How long until everything is cut together ?
Love to hear how the different cameras cut together in a live situation.
How much grading, tweaking and playing in post was needed to make them match?

cheers
Rohan

[Bob Hart]

Rohan.


My understanding is that they will try a few workflows to arrive at the best then go with that for the full thing. I also understand that they may convert everything to a common codec, do a sort of offline edit in lower res then conform the higher definition material.

It is early days yet. There are two more shows to go. I have not yet had a chance to hear playback from the full-on separate audio recording system they are using but I am told from other source that these new all digital systems are pure silk.

I'll pass this on to the people who are heading his project. I have no official status beyond shepherding one of the JVCs, so I am not a sanctioned deliverer of information.

There may be an already established line and process of publicising the project so I don't want to overstep any mark by giving any more away at this point except to say that the music and performers are magic on the nights.

When you observe audiences both cheering and clapping but two later moved to silent tears during a particular mood piece, you know the thing is working and will be a good gig when it goes to air.

[Bob Hart]

"Sonic Sessions" was convened tonight indoors at the "Fly By Night Music Club" Fremantle due to concerns about possible inclement weather which were justified.

This time an EX1 was also tossed into the camera mix for roving close-ups as the height of the stage at "The Fly" meant a handheld could stalk about without being a distraction .

Paolo's Truecolor 3 was dialed in to the three JVCs this time round. The detail on the JVCs was set back to -7 which I thought was a bit savage. Detail set at around the default level is a personal preference of mine but does not necessarily reflect best practice.

Again a welcome delight. Hank Marvin and a three and a one time four piece band when Lucky Oceans joined in. The pieces reprised were from Django Reinhart.

[Jason Rodriguez]

We will be decoupling playback and recording resolutions in future versions of SiliconDVR, but the reason why it is coupled right now is because the playback at Quadlet uses a lot more CPU . . . if you did not have enough CPU for encoding of the CineForm codec and display at Quadlet, then we've generally found that you will not have enough for decoding and Quadlet playback as well . . . which means when you go to playback, the playback is not real-time, and gets jumpy with dropped frames and odd "catch-up" moments where it trys to speed-up and "catch-up" to where the current frame should be.

So if you can record in Quadlet, you can playback in Quadlet. If you are recording in a mode and it won't work at Quadlet, then playback won't work in Quadlet either. The amount of CPU needed for decoding and encoding is approximately the same, so this is way that is the case.

Thanks,

Jason